Tuesday, June 22, 2010

Does 1/2 Marathon predict Mount Washington Time?

Updated with three years of data! The above graph illustrates the joint times of all male sub 1:45 runners at the New Bedford half-marathon that also raced the Mount Washington Road Race the same year for the years 2008-2010. The black line is a simple linear regression. The red line is a simple linear regression but the New Bedford times are log transformed with the natural log. The blue line is the line of equivalent times at both races.

1. There is a big difference between how sub 1:20 and over 1:20 half-marathoners perform at Mount Washington. Sub 1:20 half-marathoners don't seem to blow up. Over 1:20 half-marathoners do. Sub 1:20 half-marathoners often run faster at Mount Washington. Over 1:20 half-marathoners less commonly.

2. The New Bedford predictor is much steeper than the blue line, so the faster you are, the more likely you will get close to or break your half-marathon time at Mount Washington and the slower you are the less likely (a re-statement of #1).

3. The Mount Washington race predictor spreadsheet (downloadable from the race web site) predicts an expected time range, which is essentially a 5 minute window from 5 minutes less than your 1/2 marathon time up to your 1/2 marathon time. So the blue line in the graph is at the high end of the website's predictor. The New Bedford data show that this is really only a good predictor for the elite (sub 1:10 runners). For 1:10 - 1:20 half-marathoners, your half-marathon time is a pretty good predictor of your Mount Washington time.

4. If your half-marathon time is over 1:20, Mount Washington prediction is difficult. If you run a good race you should be close to your half-marathon time. But you are also more likely to blow-up (see final comment below).

5. The website predictor is optimistic if you are not an elite runner. And the slower you are, the more optimistic it is. Compare:

6. Does the official race predictor make sub-elite and especially over 1:20 runners go out too fast, sending them into a slow death spiral from which they won't recover?

Caveats - I don't have good data for sub 1:10 half-marathoners because not enough run both New Bedford and Mount Washington. Even worse, the sharp deviation of the curved line at the elite end may be heavily influenced by a single elite mountain runner, Eric Blake, whose Mount Washington Times are 7-11 minutes faster than his New Bedford times.


  1. I'm a curve breaker for sure. 1:20 at NB, Dunn-10sec at MW. The problem with NB is it takes place so early in the year. I'm definitely NOT in 1:20 shape at the moment so it's a unfair comparison for me at least. NB had decent weather this year, wasn't a factor.
    Good stuff but it isn't convincing me to run it again :-)

  2. Wolfe, will you please stop? Dogging me on my blog is one thing. Doing me on other people's blogs is something all together different. And wrong. Jeff, great regression equations. The last 1/2 I did was Big Lake, 1:25ish. I was Wolfe+10 at MWRR. Interesting.

  3. The only thing I understood in that post was "slow death spiral." I have that sucker mastered.

  4. Hey Chris, at least I got my own dot on Jeff's graph.
    I didn't use a race predictor on MW (can you tell?). I went with the deer fly strategy and pestered certain folks all the way up. Apparently the strategy is still in place.